Training using CPU

[anaghasethu]

I attempted to train the model on my X86 PC using the CPU and a dataset from Kaggle, as you suggested. Initially, the program was terminated due to high memory usage, so I reduced the batch size to 64 and increased the number of epochs to 300. However, the model's output did not meet my expectations.

Here are the steps I followed: I ran make_json.py, followed by recons_data.py, and then train.py.

Did I miss any steps, or could you help me understand why the output isn't as expected?